Multi-point shaking table test design for long tunnels under non-uniform seismic loading: Fid-Bau Portal

Multi-point shaking table test design for long tunnels under non-uniform seismic loading

Yan, Xiao / Yuan, Juyun / Yu, Haitao / Bobet, Antonio / Yuan, Yong

Highlights • A test design of a forty-meter long tunnel using four shaking tables was proposed. • The design details include similitude relation, model box, input mode and model soil. • Design of model soil was verified by single-point shaking table test and computation. • Effects of wave passage were reproduced well in multi-point shaking table test.

Abstract The design of a forty-meter long model used to estimate the seismic response of the immersed tunnel of the Hong Kong-Zhuhai-Macau (HZM) bridge engineering is presented. The model was composed of twelve model boxes, four “active” as they were placed on top of four shaking tables, each with a dimension of 4×6m with capacity of 1.5g. The remaining eight model boxes were “inactive”, as their response was the result of the motions imposed by the active model boxes through the connection between boxes. The design of the connections was deemed critical to accomplish a consistent excitation through the system. Analytical and numerical simulations of the test system were carried out where the model boxes were represented as beams and their connections as rotational springs. In the model, wave passage along the axis of the tunnel was input to the “active” model boxes using a delay equal to the time it would take the seismic wave to travel from one active box to the next along the axis of the tunnel. A series of tests conducted with the boxes verified the design. The in-situ soil, found at the location of the tunnel, was modeled using a synthetic soil formed by an intimate mixture of sawdust and sand with mass proportions 1:2.5, that yielded dynamic properties that, after the similitude ratios used for the design of the system, were analogous to those of the natural soil. The effects of the rigid boundaries imposed by the lateral walls of the model box were investigated by carrying out tests on a single model box, filled with the synthetic soil that reproduced the natural slope of the ground at the location of the tunnel. Readings from accelerometers placed inside the soil during the shaking table tests showed that the effects of the boundaries were negligible. These results were confirmed by two-dimensional plane-strain simulations using a finite element method that replicated the experiments. In summary, analytical, numerical and experiment results consistently showed that the design of the model boxes and the synthetic soil were adequate. The design and tests of the long tunnel model will be presented elsewhere.